TiPPLNS stands for Tier-weighted PPLNS. It's our normal PPLNS payout with one addition: it honors the reward structure that the Keryx network itself defines for AI models.

Keryx is more than proof-of-work — every block also proves an AI model ran on your GPU (Optimistic Proof of Inference). The network pays a higher block reward for heavier models, because they do more valuable work. TiPPLNS simply passes that network reward through to the miner who earned it. Nothing is taken from anyone: you receive exactly what the network paid for the blocks you helped find.

Your payout flows through two layers:

1. Normal PPLNS — you're rewarded for the work (shares) you contribute, exactly as you always have been.
2. Network reward pass-through — that reward is matched to the model tier of the blocks, so the extra value Keryx assigns to bigger models flows to the miners running them.

It is revenue-neutral for the pool — we don't add a fee or hold anything back for this. We pass through the on-chain reward structure the network already created.

Some miners ask us to open a separate pool for each AI model — one for the light model, one for the heavy model, and so on. You don't need that, and it would actually pay you less. Here's why.

On Keryx the model tier is baked into the block reward itself — it is not a pool-side accounting trick. When your GPU mines a heavier model, the network pays a bigger reward for the blocks you find, and TiPPLNS routes that exact bonus straight back to you. So a miner on the base model and a miner on the top model can sit on the same pool at the same time and each still earns precisely what their own model is worth — base pay for base, +67% for the top tier.

Both miners above are in one pool, contributing to the same combined hashrate — yet each is paid at its own model's rate. That is the “pool-in-pool”: every model tier effectively has its own reward rate inside the single pool, with nothing to configure and no hashrate to split.

A separate pool per model would only make things worse:

• It splits the hashrate. Fewer miners per pool means each pool finds blocks less often, so your payouts get lumpier and swingier (higher variance/luck). Staying together keeps everyone's payouts steady.
• It gains you nothing. The bonus is already applied per miner, per block, by the network — splitting the pool can't increase it. You'd take on all the downside for zero upside.
• It's more to maintain (separate stratum, stats, payouts) with no benefit to a single miner.

Running a smaller model is completely fine — you always earn the full network reward for that model, and your pay for the work you do is unchanged. Upgrading to a larger model is simply the way to unlock the higher rewards Keryx already offers for them.

On Your Stats you'll find a Model column on every round and a Model Bonus summary showing the network reward your model earned you — plus a live Pool Model Mix of what the whole pool is running.